Abstract:
This study aims to use the survival methods in large censored data, sample of 1098 Sudanese children under five years from both categories, affected with different five diseases are acute renal failure, congenital deformity heart, leukemia, septicemia and sickle cell disease) at Jafar Ibn Auf Pediatric Hospital in Khartoum/ Sudan,from 2012-2016. The purpose is to estimate the accurate probability of survival time for children in existence heavy censoreddata. The two methods used are Modified Weighted Kaplan-Meier (MWKM) and Accelerated Failure Time (AFT) model using SPSS, STATA, NCSS and XLstat. The main hypotheses tested arethat no differences between probability survival and hazard rates, Cox proportional hazardmodel and accelerated failure time model in estimating the probability survivaltime. The results obtained 235(21%) of children diedduring the study with median’s survival timeof 16 day per disease and 863 (79%) are censored till the study end with survival rate (0.97). Modified Weighted Kaplan Meier estimator gives accurate survival probability of 100% to the last censored child, and when the survival of Child (j) is 𝑝𝑗 equal 0.1 to 0.9, the estimator gives the accurate probability survival time to the last censored childrenfrom 0.2167 to 0.9002, respectively. Moreover, AFT model is better than Proportional Hazard model in estimatingthe large censoreddata. This proofed through goodness-of-fit test (AIC&BIC), which obtained Weibull AFT model has fitted better, more valuable and realistic predicted thatthe survival and hazard functions than Proportional Hazard model. The study recommended using MWKM and AFT model in estimating the probability survival timeforsuchdataset.
